CodeVista: A Vital Tool for Eager Testers

Curious about how CodeVista operates and eager to compare its performance with ChatGPT and OpenAI, Test Lead Đặng Hải Nguyên (CEP.ESS) found that after some time using it, CodeVista has become an indispensable assistant in his daily work, particularly in developing coding and scripting skills.

Đặng Hải Nguyên, currently a Test Lead for Microservices-related products in the Oil & Gas sector at CEP, began using CodeVista in mid-August. This initiative was part of a campaign kick-off and leadership call at FSU, aimed at supporting self-learning in Automation Testing on mobile apps. He explored how CodeVista handles commands and its response times as an in-house AI tool.

Starting as a Manual Tester, Nguyên faced challenges due to a lack of TechStack knowledge, which made getting accustomed to coding time-consuming. Forums often failed to provide effective solutions. “CodeVista has helped me solve all these issues during my self-learning process,” Nguyên stated.

“The answers, interspersed with sample code and annotations, make everything clearer and easier to understand. The new code snippets suggested by CodeVista are more concise and help me avoid overusing conditional statements or loops, which I previously struggled with,” he added. Additionally, CodeVista can be integrated into Visual Studio Code IDE as an extension, allowing users to work quickly and conveniently, significantly boosting productivity.

From his personal experience, Nguyên shared that the key to effective commands lies in using “concise” and “succinct” keywords. “For complex issues, I break them down into multiple questions. Each question includes the TechStack keyword I need. This sequential questioning helps the AI understand the problem more deeply and connect all the events to provide a more focused answer instead of rambling,” he explained. Breaking down requests also gives users the feeling of being heard by a rational entity, rather than just receiving pre-prepared answers like a chatbot.

During his usage, Nguyên also identified areas for improvement in CodeVista, such as server management and daily data quotas. He explained that because CodeVista is managed on FPT’s server and network, it cannot be deployed for projects that require working within the client’s network. Additionally, the daily data quota is limited to 100 commands, which may be uncomfortable for users needing to input extensive data to solve complex cases.